526 EHRLICH, McGOWEN, AND MUSZYNSKI 



uted along the length of the test chamber in the following pattern: 

 21 on the left, 12 in the center, and 20 on the right. The water at 

 these times was at the fish's acclimation temperatures. A similar 

 attraction for tank ends was noted by Doudoroff (1938) for 

 atherines of undesignated species. As soon as a thermal gradient was 

 established, the fish moved toward the warm end of the tank. They 

 were then distributed in the following pattern: 1 in the cold portion, 

 10 in the middle, and 42 in the hot portion of the gradient. The 

 differences in fish distribution before and after establishment of a 

 temperature differential were observed just before the change in 

 temperatures began and 15 min later. 



Although the fish moved freely throughout the experimental 

 chamber when it was at a uniform temperature, after the formation 

 of a thermal gradient, the topsmelt became positioned in the 

 compartment they initially selected; this was generally in the 

 temperature range of 19 to 23° C. Shifting isotherm positions did not 

 cause the fish to move until they contacted temperatures of 26 to 

 27°C, at which time they started tracking this temperature range and 

 avoiding warmer waters. This behavioral pattern was shown by 

 18-day-old larvae and by all juveniles tested (Figs. 1 and 2). Younger 

 fish showed less discrimination with respect to thermal avoidance 

 (Fig. 1). Doudoroff (1945) found juvenile topsmelt to be eury- 

 thermal (96-hr viable temperature range, ~10 to Sl^'C). Hubbs 

 (1965) found a slightly narrower range during the embryonic stages. 

 It may be that, for species with a wide thermal tolerance, as Brett 

 (1956) suggested for salmonids, temperature is not a strong directing 

 force until the fish's lethal range is approached. The upper viable 

 limit for hatching of topsmelt eggs is 26.8°C (Hubbs, 1965). 

 Assessment of temperatures allowing development may give a more 

 sensitive indication of the viable range than simply placement of 

 juveniles in water baths at different temperatures for 4 days. 

 Apparently, in the terminology of Reynolds (1977), temperature for 

 this species should be considered a proximate rather than an ultimate 

 factor in habitat selection. Topsmelt show a stronger avoidance than 

 preference response. Beitinger (1977) showed that little variation in 

 avoidance of water near lethal temperatures was tolerated by bluegill 

 sunfish (Lepomis macrochirus). The relationship between selected 

 temperature and time in the gradient is shown in Figs. 1 and 2 for fish 

 at 11, 18, and 192 days post-hatching. The two younger groups were 

 larvae. Comparison of the results from these two groups shows that 

 by 18 days the larvae started to show the behavioral patterns of 

 juveniles, i.e., avoidance of water warmer than 27°C and aggregation 

 of individuals within the gradient. The tendency of older larvae and 



